1. Field
Embodiments of the present invention relate to a technique of locating an abnormal point of a network.
2. Description of Related Art
As an electrical data network, such as a computer network, becomes complicated and diversified, a technique of locating an abnormal point of the electrical data network (hereinafter “network” for short) quickly and accurately is sought after in the field of network monitoring/failure monitoring.
Conventionally, as an example of a technique of locating an abnormal point of a network, a network tomography analysis technique of analyzing an abnormal point in a network based on normality/abnormality information of an end-end observation flow (measurement flow) in the network is known (for example, see Japanese Laid-open Patent Publication No. 2006-238052).
Moreover, as another example of a technique of locating an abnormal point of a network, a technique of making a network abnormality diagnosis by using a tree structure is known (for example, see Japanese Laid-open Patent Publication No. 2006-229421).
For example, an outline of the network tomography analysis technique is shown in FIGS. 1 and 2. FIG. 1 illustrates an example of a network that locates an abnormal point. Moreover, FIG. 2 is an example of a matrix in which an observation flow passage link through which an observation flow in which an abnormality has been observed passes (goes) is mapped to an observation flow.
The network shown in FIG. 1 has flow quality measurement agents A to E, relay nodes R1 to R6, and links 1 to 12. Incidentally, the links 3 and 4 are abnormal links (links in which packet loss occurs) in which an abnormality occurs. Moreover, an observation flow is represented by a symbol for a source flow quality measurement agent and a symbol for a destination flow quality measurement agent. For example, an observation flow observed by a source flow quality measurement agent A and a destination flow quality measurement agent B is represented as an observation flow AB.
For example, a general procedure of the network tomography analysis technique includes procedure 1 by which a judgment (or determination) is made as to whether each observation flow is normal or abnormal, and normality/abnormality information is mapped to a link (an observation flow passage link) through which the observation flow passes, and procedure 2 by which a link set covering an observation flow (an abnormal flow) in which an abnormality has been observed in a minimum number of links is judged to be a suspected point. In the case of the matrix of FIG. 2, since two links (a minimum number), which are the links 3 and 4, can cover all abnormal flows, the links 3 and 4 are identified as an abnormal point.